--- license: mit language: - rw base_model: - Davlan/afro-xlmr-base --- ### Distill Afro-XLMR Base Base Model (12 Layers): Embedding Dimension: 768 Number of Layers: 12 Total Parameters: 278,295,186 Estimated Size: 1061.61 MB Reduced Model (4 Layers) Embedding Dimension: 768 Number of Layers: 4 Total Parameters: 221,592,210 Estimated Size: 845.31 MB Parameter Reduction from Base Model (12 Layers) to Reduced Model (4 Layers): 20.38% Size Reduction from Base Model (12 Layers) to Reduced Model (4 Layers): 216.30 MB